Mangrove Mapping with Satellite Imagery: Protecting Coastal Forests from Space

Standing in a mangrove forest in southern Thailand, ankle-deep in brackish water with aerial roots arching overhead, I remember thinking: this is the most difficult ecosystem to survey on foot. Dense, muddy, tidal, full of mosquitoes. A 100-hectare mangrove stand might take a field team a week to traverse. A satellite covers it in milliseconds.

Mangrove mapping from space isn't just convenient — for many remote coastlines, it's the only practical option.

Why Mangroves Are Spectrally Distinctive

Mangroves have several properties that distinguish them from other vegetation in satellite imagery:

Waterlogged substrate: The intertidal soil/water beneath the canopy affects the spectral signal, particularly in SAR where the water surface creates a double-bounce reflection.

Canopy structure: Mangrove canopies are typically dense but relatively short (5-25 m), with a different structure than terrestrial forests.

Species composition: Mangrove species have different leaf properties than most terrestrial vegetation — waxy, thick leaves adapted to salt tolerance.

Location: By definition, mangroves occupy the coastal intertidal zone. This geographic constraint is a powerful classification aid — vegetation in the intertidal zone is very likely mangrove.

Optical Mapping Approaches

NDVI + Elevation/Location

The simplest approach:

1. Compute NDVI from Sentinel-2
2. Apply a vegetation threshold (NDVI > 0.3)
3. Restrict to coastal zones within elevation range (0-10 m above sea level, from a DEM)
4. Exclude known non-mangrove vegetation types

This works surprisingly well in many regions because few other vegetation types occupy the low-lying coastal zone.

Spectral Indices for Mangrove Discrimination

CMRI (Combined Mangrove Recognition Index): CMRI = NDVI − NDWI

This index exploits the fact that mangroves have high NDVI (like other forests) but also higher NDWI (due to the wet substrate), giving them a distinctive combined signature.

SWIR-based separation: Mangrove soil is typically wet/saturated, producing lower SWIR reflectance compared to terrestrial forests on dry soil. The ratio of NIR to SWIR helps separate the two.

Multi-Temporal Analysis

Mangroves are evergreen — they maintain green canopy year-round. In regions where adjacent terrestrial vegetation is seasonal (deciduous forests, seasonal crops), a dry-season image clearly distinguishes evergreen mangroves from senesced/bare surroundings.

SAR for Mangrove Mapping

SAR has unique advantages for mangrove monitoring:

Double-Bounce Mechanism

Radar signals hitting the water surface beneath mangrove canopy bounce off the water, reflect off tree trunks, and return to the satellite. This double-bounce produces a strong, characteristic signal in HH polarization (horizontal transmit, horizontal receive).

For Sentinel-1 (VV/VH polarization), the VH channel captures volume scattering from the canopy, while VV includes both surface and trunk-ground interactions. The combination provides reasonable mangrove discrimination.

Cloud Independence

Tropical coasts where mangroves grow are among the cloudiest places on Earth. Optical sensors may not acquire a usable image for weeks. SAR monitors continuously regardless of weather — essential for near-real-time deforestation detection.

Tidal Considerations

Water level beneath the canopy varies with tides. At high tide, more water is present, enhancing the double-bounce signal. At low tide, exposed mudflats reduce the double-bounce. SAR-based mangrove mapping should account for tidal state at the time of acquisition.

Global Mangrove Watch

The Global Mangrove Watch (GMW) dataset provides:

• Global mangrove extent at ~25 m resolution
• Time series: 1996, 2007, 2008, 2009, 2010, 2015, 2016, 2017, 2018, 2019, 2020
• Method: Combination of ALOS PALSAR (L-band SAR) and Landsat optical data
• Classification: Random Forest trained on known mangrove locations

The time series reveals global trends: total mangrove area decreased from approximately 142,866 km² in 1996 to about 136,000 km² in 2020 — a net loss of approximately 5%, despite some areas showing recovery or expansion.

Why Mangrove Loss Matters

Mangroves provide ecosystem services far exceeding their spatial extent:

Carbon storage: Mangrove soils store 3-5× more carbon per hectare than terrestrial forests. When mangroves are destroyed, this "blue carbon" is released, contributing to greenhouse gas emissions.

Coastal protection: Mangrove roots dissipate wave energy, reducing coastal erosion and storm surge damage. Studies show mangroves can reduce wave height by 66% over 100 m of forest width.

Fisheries: Mangrove ecosystems serve as nursery habitats for commercially important fish and shrimp species. Loss of mangroves directly impacts coastal fisheries productivity.

Biodiversity: Unique species assemblages adapted to the brackish intertidal environment.

Monitoring Loss and Recovery

Deforestation Detection

Mangrove loss is detected similarly to terrestrial deforestation — NDVI drops, SAR backscatter changes — but with additional considerations:

• Aquaculture conversion: The most common driver of mangrove loss (especially shrimp ponds). Regular geometric shapes appearing in mangrove zones indicate aquaculture expansion.
• Coastal development: Urban/port expansion into mangrove areas.
• Natural dieback: Storm damage, sediment starvation, sea level rise.

Restoration Monitoring

Mangrove restoration projects (planting or natural regeneration) can be tracked from satellite:

• Early stage: Seedlings too small to detect spectrally, but changes in tidal flat reflectance may be visible
• Canopy development (2-5 years): Increasing NDVI as canopy closes
• Mature restoration (5-15 years): Spectral signature approaches natural mangrove; SAR double-bounce develops as trunks grow

Satellite monitoring provides independent verification of restoration success — important for projects receiving carbon credit financing.

Challenges

Mixed pixels: Mangrove boundaries are often gradational — dense mangrove transitioning to sparse mangrove to tidal flat to water. At 10-20 m resolution, boundary pixels contain mixtures.

Species-level mapping: Different mangrove species have subtly different spectral signatures, but reliable species discrimination from Sentinel-2 alone is difficult. Hyperspectral sensors or very high-resolution data can improve this.

Shoreline dynamics: Mangrove positions shift with sediment accretion and erosion. What appears as "mangrove loss" in one location may be compensated by "mangrove gain" elsewhere along the coast — natural coastal dynamics rather than anthropogenic destruction.

Definition consistency: Different mangrove area estimates vary by 10-20% depending on the classification method, minimum mapping unit, and whether degraded/sparse mangrove is included.

Despite these challenges, satellite-based mangrove monitoring has fundamentally improved our ability to track these critical ecosystems. The Global Mangrove Watch and similar initiatives provide the spatial evidence needed to enforce protection, prioritize conservation, and measure the effectiveness of restoration — at scales that ground surveys simply cannot achieve.